Incorporating Patient Complexity into Physician Workload Assessment Abstract Anticipated Impact on Veterans'Healthcare Assessing the likelihood of work to be generated by a given patient could allow for the safer allocation of patients among admitting physicians. This is especially important in teaching Veterans Affairs Medical Centers (VAMCs), but is also applicable to those using hospitalists. The anticipated impact on veterans is a reduction in adverse events associated with mistakes due to high physician workload. Background More than 500,000 hospitalizations occur yearly in the VAMC system. Inpatient physician workload is a potential contributing factor to patient adverse events. The current method for approximating inpatient physician workload is census. This is thought to be a poor estimate of workload because it does not take into account any patient factors, such as acuity and comorbidities. It is common nursing practice to generate an acuity score for individual patients in order to allocate work. Drawing on nursing practice and human factors engineering, we propose a pilot study to establish a method for collecting data on physician workload as it relates to individual patients. Objectives 1) To establish a method for measuring the workload generated by individual patients. This will be accomplished by directly observing physicians as they admit patients, and then recording the amount of time that physicians spend on individual patients;and 2) To explore the ability of several factors to predict the amount of physician time spent on individual patients during the first 12 hours of admission. Methods We will recruit 20 physicians at the Milwaukee VAMC. Data collection will occur when the physicians are admitting patients. Direct observation will be used to perform detailed task analysis. This will result in a record of the amount of time spent on individual patients. The data collection will be iterative allowing for adjustment to the methods as the data collection proceeds. Retrospective chart review will occur for two patients admitted by each participating physician. The chart review will include demographics and other data needed to complete standardized instruments assessing acuity. Analysis will include calculating descriptive statistics for the amount of time spent on individual patients in the first 12 hours after admission. We also plan to calculate agreement for the task analysis on a subset of observations. We will analyze data collected from patient chart review and their correlations with the amount of time spent by the physician on the individual patients. This work will result in the establishment of a method to collect the data needed for task analysis of physicians with respect to individual patients. It will also provide preliminary data for a larger multi-center study to identify a prospective method for predicting the likely workload associated with individual patients.